摘要
在经典算法中由频繁项集生成关联规则需要生成频繁项集的所有非空子集作为候选后件集。李雄飞对此做出改进,提出逐层搜索后件的宽度优先算法。求下集极大元的Boundary算法也可用于求所有关联规则后件。论文提出一个深度优先算法GRSET(GenerateRulesbyusingSet-EnumerationTree),该算法利用集合枚举树,按照深度优先的方法逐一找出所有关联规则后件并得到相应的关联规则。通过实验对这三种算法进行比较,结果显示GRSET算法效率较高。
The classical algorithm of mining association rules gnerated by a frequent itemset has to generate all nonempty subsets of the frequent itemset as candidate set of consequences,Li Xiongfei aimed at this and proposed an improved algorithm.The algorithm finds all consequences layer by layer,so it is breadth-first.We also can use Boundary algorithm of finding all maximal elements of a lower segment to get all consequences of the association rules,ln this paper,we propose a new algorithm GRSET(Generate Rules by using Set-Enumeration Tree) which uses the structure of Set-Enumeration Tree and depth-first method to find all consequences of the association rules one by one and get all association rules corresponding to the consequences.Experiments show that GRSET algorithm is more efficient than the other two algorithms.
出处
《计算机工程与应用》
CSCD
北大核心
2006年第26期152-155,共4页
Computer Engineering and Applications
基金
国家自然科学基金资助项目(编号:60474022)
河南省骨干教师资助项目(编号:G2002026)
河南省自然科学计划资助项目(编号:200510475028)
关键词
数据挖掘
频繁项集
关联规则
深度优先算法
data mining,frequent itemset,association rules,depth-first algorithm